After the Hike, Who Is Actually Setting the Gold Price?
This article is for educational purposes only and does not constitute financial advice. Do your own research before making any investment decision.
Nine months ago the market expected two Fed cuts in 2026 and crude traded near $57 a barrel. On September 17 crude was still above $100, the Fed had raised rates by 25bp, and the market was pricing two more hikes before year-end.
Between those two worlds gold rallied, peaked, fell, rebounded and gave back part of the rebound. It did not stop working as a hedge, and it did not decouple from real yields and the dollar.
Gold answers to four forces at once: real yields, the dollar, hedging and inflation demand, and positioning plus the structural bid. In 2026 those four stopped pointing the same way.
Read gold on three clocks. Days and weeks belong to flows. Six months belong to rates and the dollar. Years belong to whether central banks still treat gold as a reserve asset.
Four Ledgers, Not One Formula
Gold pays no coupon and earns no profit. Your return comes from the price, so the higher the real return available on cash and Treasuries, the more it costs to own an ounce instead. A stronger dollar usually weighs on the price too, because buyers outside the United States pay more in their own currency.
That framework matters and it is incomplete. Gold also responds to hedging demand, central bank purchases, ETF creations and redemptions, futures positioning and forced deleveraging. In any given week real yields can push the price down while central bank and ETF buying holds it up.
Think of the price as four separate ledgers:
- Opportunity cost. Real yields set what you give up by owning gold instead of a risk-free asset.
- Currency. The dollar sets the real purchasing power of every non-dollar buyer.
- Insurance. Geopolitical risk and inflation expectations set how many people need the hedge.
- Positioning and the bid. Futures leverage, ETF flows and reserve allocation decide who is actually transacting.
The difficulty in 2026 is not that these relationships broke. It is that they stopped agreeing. Higher oil raised both hedging demand and rate hike expectations. Central banks kept buying into weakness while Western ETFs redeemed through the second quarter. The dollar softened at times, yet long yields stayed elevated on Treasury supply, inflation and competing demand for capital.
January: A Record High, Then a Margin Call
Gold rose roughly 64% in 2025. Entering 2026 the market still expected two cuts, and geopolitical risk plus the US debt trajectory were both working in gold’s favour.
On January 28, 2026, gold set an all-time high of $5,589.
The rally had also built up its own fragility. The 2025 to January 2026 move in precious metals pulled in retail money, leveraged ETFs and trend followers. Once the price turned, daily ETF rebalancing, futures margin and stop losses amplified the selling in the same direction.
So the first leg down in late January was a positioning event, not a macro event. The macro deteriorated afterwards, and that turned a technical flush into a six month drawdown.
How $100 Oil Rewrote the Rate Path
Crude started the year near $57. By mid September it was above $100.
Gold is usually described as an inflation hedge, which makes rising oil sound bullish. In 2026 the transmission ran the other way:
- Crude lifted gasoline, freight and production costs.
- Disinflation stalled.
- The market first removed cuts, then priced hikes.
- Cash and short-dated yields rose, supporting the dollar.
- Gold, which pays nothing, absorbed the higher opportunity cost.
That is why war and an energy shock did not produce a sustained one-way safe haven bid. Precious metals kept falling even as the Middle East conflict escalated, which cuts against the reflex that conflict is automatically good for gold.
In January the market traded cheap oil, cuts and a softer dollar. In September it traded expensive oil, hikes and rates staying high. This was not an ordinary pullback within one macro regime. The pricing basis itself changed.
February to July: Two Buyers on Different Clocks
The January flush cleared the positioning problem and did nothing about oil or rates. Through the second quarter, North American investors marked up their inflation and rate expectations, and gold ETFs shed 45 tonnes. Over the same period central banks bought 289 tonnes net, up 62% year on year.
These two pools are solving different problems. ETF investors care about the next Fed meeting, the dollar and the cost of carry. Central banks care about reserve diversification, sanctions risk, geopolitical exposure and the shape of the monetary system several years out.
That is how gold could fall without losing its floor. Continuous futures bottomed near $3,980 on July 17, about 28.8% below the January high.
August: A Rebound Built on Flows, Not Cuts
Gold gained roughly 13% in August and finished the month near $4,563, its third-largest monthly advance in 25 years. ETF buying and futures money did most of the work, with a softer dollar providing a second layer of support.
None of that meant the easing cycle was back. It looked much more like a rebound off a low base, driven by positioning and returning flows. Continuous futures reached about $4,716 on August 26, still 15.6% below the January high.
In early September gold briefly rose even as the odds of a hike climbed. That divergence shows short-term positioning and ETF flows can overwhelm the day’s rate signal. It does not show that rates stopped mattering. Once the Fed confirmed the path at its meeting, gold turned back down.
September: One Hour Did All the Damage
On September 16 the Fed raised rates 25bp to a 3.75% to 4.00% range. On the dot plot, 16 of 18 officials saw more tightening ahead. The market then moved to price two more hikes by year-end.
Conventionally that is bearish for gold. Higher rates mean a higher cost of holding an asset that yields nothing.
The dashed red line on the chart is the moment of the announcement, 18:00 UTC, 2pm in New York. That line is the only reference point you need.
To the left of it, gold spent more than a day inside a range less than $50 wide. Nobody wanted directional risk before the result. In the final hours before the line, gold actually drifted higher, touching $4,359.
The first segment to the right of the line is the cliff. Within one hour gold went from $4,359 to $4,261, a fall of $98, or 2.25%.
That hour contains the entire decline. Gold was rising before it and gold was rising after it. The whole loss was compressed into sixty minutes.
This is where the popular explanation fails. If the hike had genuinely been discounted in advance, that hour would not have cost $98. The market only received the dot plot in that hour, and only then learned that 16 of 18 officials were not finished.
What happened was a positioning event, not a repricing of value. Ahead of the decision uncertainty was high and holders cut exposure. Once the result was public the uncertainty disappeared, shorts pressed first, then covered. The advantage of this explanation over “already priced in” is that it leaves evidence: you can check it against futures positioning and ETF flows.
The selling stopped inside that hour. Gold spent the rest of the session flat around $4,264 without making a new low. The next day it climbed back to $4,373, recovering $112, or 2.6%, more than it had lost. By the right edge of the chart gold sits above the pre-decision range, $12 short of the high it printed before the cliff.
The explanation is on the oil panel below.
WTI peaked at $106 to the left of the red line, a full day before the Fed. It then fell for three consecutive sessions: down 3.2% to $102 on September 16, briefly under $100 intraday on September 17, settling at $102. From the high to that settlement, a 4.3% decline over three days.
The trigger was specific and had nothing to do with the Fed. Saudi Arabia began rerouting crude through Oman and the damaged East-West pipeline was due back online, which took the edge off supply disruption fears.
Lay the two panels on top of each other and gold’s climb sits directly over the stretch where oil worked its way toward $100.
The case for hiking was inflation driven by energy. Three sessions of falling crude, roughly 4% in total, removed part of that case. The hike raised gold’s cost of carry while cheaper oil lowered both inflation and hedging demand. Two forces pulling opposite ways, and on September 17 the second one eased first.
Magnitude is the part worth being precise about. Gold’s daily correlation with oil this year is only -0.09, and even across these two days the hourly correlation is just -0.25. Small moves in crude do nothing to gold because the chain of transmission is too long. This move was visible because it ran three sessions in a row and carried a supply story anyone could read.
Ranking the Three Usual Suspects
Three variables get used to explain gold. Their track records this year are not close. The figures below are correlations, bounded between -1 and 1. Negative means the two move in opposite directions, a larger absolute value means a tighter link, and a value near zero means almost no relationship.
Daily data, January to September: the dollar at -0.50, the 10-year real yield at -0.29, WTI crude at -0.09.
The dollar is the only dependable signal of the three. Dollar up, gold down, and the reverse.
Oil has almost no direct relationship. Crude nearly doubled this year, from about $60 to about $100 a barrel, while gold chopped sideways.
Oil’s influence is indirect: it moves inflation, inflation moves policy expectations, and expectations move gold. Every link in that chain absorbs part of the signal.
The real yield is the nominal yield minus expected inflation, which is the cleanest measure of gold’s cost of carry. Textbook theory wants a strong inverse relationship. This year it is -0.29, visibly weaker than advertised.
Why One Month of Data Proves Nothing
A common claim is that gold’s link to real rates has broken down or even inverted, based on a one-month correlation of -0.39 against -0.29 for the full year.
With a small sample, a number like that is mostly coincidence. What tells you whether to trust it is the confidence interval, the range the true value plausibly occupies. Narrow is credible.
One month is 21 trading days, and the intervals are too wide to be useful. For the real yield the one-month interval runs from -0.70 to +0.05, wide enough to contain both a strong inverse relationship and no relationship at all. For oil it runs from -0.63 to +0.17, also straddling zero.
Test whether the recent window genuinely differs from the full year and the dollar returns a p-value of 0.105, the real yield 0.648. By any conventional standard, neither comes close to a real difference.
The breakdown story simply is not supported by 21 observations. The full-year figures are what you can use, and the dollar’s -0.50 is the only one whose interval stays clear of zero.
Rising Real Yields Are Not Automatically Bearish
The 10-year real yield began the year at 1.94% and reached 2.68% on September 16, up 74bp. Over the eight sessions from September 8 to 16 alone it added 25bp.
Conventionally that should suppress gold. Both rose together, which looks like a contradiction.
The reason for the rise matters, not the rise itself. There are two possibilities, and they point opposite ways:
- Fiscal. The government is issuing more debt than the market wants to absorb, so investors demand more compensation to hold duration. Here the rise reflects sovereign risk, and gold, which hedges exactly that, need not fall.
- Growth and credibility. The economy is strong and investors believe inflation will be contained. Here the rise is unambiguously bearish for gold.
One spread tells you which regime you are in:
The 10-year nominal yield minus the 10-year real yield. That difference is the market’s expected long-run inflation rate, known as breakeven inflation. If it widens, the market is worried about inflation. If it stays flat while real yields climb, the market is charging a fiscal risk premium.
This year’s numbers settle it. The nominal yield went from 4.19% to 5.01%, up 82bp. The real yield went from 1.94% to 2.68%, up 74bp. The spread moved from 2.25% to 2.33%, up 8bp.
Roughly 90% of the rise in long-term yields came from real yields. Inflation expectations barely moved. All year the spread stayed locked between 2.17% and 2.50%, a band of just 33bp.
The last month is more extreme still. Over 21 sessions the real yield added 27bp and breakeven inflation added 3bp.
This is the fiscal case, and not a borderline version of it.
The market is not pricing inflation. It is charging more to hold US duration. American deficits stacked on top of enormous AI-related financing are draining global long-term savings and raising the true price of capital.
What gold is pricing here is not its cost of carry. It is that fiscal pressure, which is another way of saying it is pricing a weaker dollar.
The Bid That Does Not Watch the Screen
Underneath the noise sits a group of buyers who are indifferent to the price.
Global gold ETFs took in about $17.9bn in August, adding 121.2 tonnes and lifting total holdings to a record 4,189.2 tonnes. North America contributed $7.7bn and Europe $7.9bn, its strongest month on record. China has now added to its gold reserves for many consecutive months.
Central banks buy against a reserve allocation target, not against last week’s price. ETF money is there because of concerns about fiscal sustainability and currency risk, and it does not leave because real yields rose a few basis points.
This bid sets the height of the floor. It does not set short-term direction. It is why the $98 hour did not extend into a rout, and why the selling stopped near $4,261.
Put the dollar, the real yield and oil together and they explain roughly 30% of gold’s daily variance. The other 70% belongs largely to these structural buyers, who do not act on daily data and therefore cannot be captured by any daily model.
Three Things to Watch
- Watch the dollar. The strongest and most stable signal of the year at -0.50, and the only one of the three you can lean on. Dollar down, gold up, and the reverse.
- Watch the bid. Central banks and ETFs determine the floor, not the direction. Gold stops falling where someone is willing to absorb it, and on September 16 that level was $4,261.
- Watch the spread between nominal and real yields. It has moved 8bp all year, which says the rise in long yields is fiscal risk rather than inflation, and that supports gold. The day that spread starts widening in earnest, the support changes character: gold would then be trading inflation, not sovereign risk.
Oil and any single rate decision are not worth watching closely. They matter, but only sustained moves with an identifiable cause make it through the chain.
Flows in the Short Run, Trust in the Long Run
Everything above explains short-term moves. Oil fell 4%, so gold recovered. The dollar softened, so the cost of carry fell. One hour cost $98, so positioning got flushed.
All of that is path, not direction.
Direction is set by something slower: how many institutions still want to hold dollars as reserves.
Most people remember what happened in 2022. After Russia invaded Ukraine, the West froze Russia’s foreign exchange reserves. What central banks saw in that moment was not the sanction itself. It was a question they had never seriously priced: reserves held on someone else’s ledger can be switched off.
Gold did not spike that year. What changed was the volume of buying. Over the four years since, central banks have bought roughly 1,000 tonnes a year, double the 500 tonne average of the preceding decade.
Four years on, the balance sheet shows the result. At the end of 2025 gold accounted for 27% of global official reserves, against 22% for US Treasuries. For the first time since 1996, central banks hold more gold than they hold US government debt.
On what comes next, the survey data is more direct than any commentary. Among central banks, 74% expect the dollar’s share of reserves to be lower in five years, 84% expect gold’s share to be higher, and 89% expect official gold reserves to keep rising over the next twelve months. In a separate survey, for the first time, more central banks plan to reduce dollar exposure than to add it.
These are not forecasts. They are stated intentions from the people who hold the assets. The buyers are already queued up.
The dollar’s problem is not its level, it is trust. A weaker exchange rate is not the threat, and Washington is not opposed to a softer dollar. What it needs to prevent is capital flight. Stablecoin legislation and the build-out of dollar digital infrastructure serve the same purpose: hold on to the dollar’s use cases while the exchange rate drifts lower. That defends the dollar as a means of payment. It does nothing for the dollar as a store of reserves, because reserve managers are asking whether an asset can be frozen, not whether it is convenient.
Europe is building its own defence industry, the BRICS are negotiating their own settlement arrangements, and China keeps adding gold. These decisions are independent of one another and they converge on the same conclusion.
One qualification matters. This is a shift measured in years, not months. The dollar is not collapsing. In the first quarter of 2026 its share of allocated reserves actually rose to 57.13%, and most reserve managers describe de-dollarization as gradual.
So the conclusion is higher, with volatility.
The direction is settled and the path is rough. This year already demonstrated how rough. Gold peaked at $5,589 on January 28 and was still about 22% below that level on September 17. That drawdown refutes none of the structural logic. It shows that inside a hiking cycle, opportunity cost is enough to sit on the price for half a year.
The second half remains on a tightening path, with two more hikes priced by year-end. That environment rarely produces a clean one-way advance. The likelier shape is a rising range: each shock absorbed by the structural bid, each low a little higher than the last, with the ceiling waiting on a catalyst.
The catalyst can be confirmation that the hiking cycle is over, the start of a genuine downtrend in the dollar, or one more event that reminds governments their reserves can be confiscated.
One reference point worth remembering:
Across 60 central banks, the average forecast for gold at the end of 2026 is $5,354. That is the buyers quoting their own price, not a sell-side target.
Today’s price is set by this month’s oil and this meeting’s dot plot. The direction is set by something that has already happened and cannot be reversed.
点赞
收藏
点踩
分享